index/extract.c

   1 /* $Id: extract.c,v 1.278 2008-01-26 15:32:51 adam Exp $
   2    Copyright (C) 1995-2007
   3    Index Data ApS
   4
   5 This file is part of the Zebra server.
   6
   7 Zebra is free software; you can redistribute it and/or modify it under
   8 the terms of the GNU General Public License as published by the Free
   9 Software Foundation; either version 2, or (at your option) any later
  10 version.
  11
  12 Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
  13 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  14 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  15 for more details.
  16
  17 You should have received a copy of the GNU General Public License
  18 along with this program; if not, write to the Free Software
  19 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  20
  21 */
  22
  23 /** \file
  24     \brief indexes records and extract tokens for indexing and sorting
  25 */
  26
  27 #include <stdio.h>
  28 #include <assert.h>
  29 #include <ctype.h>
  30 #ifdef WIN32
  31 #include <io.h>
  32 #endif
  33 #if HAVE_UNISTD_H
  34 #include <unistd.h>
  35 #endif
  36 #include <fcntl.h>
  37
  38
  39 #include "index.h"
  40 #include "orddict.h"
  41 #include <direntz.h>
  42 #include <charmap.h>
  43 #include <yaz/snprintf.h>
  44
  45 static int log_level_extract = 0;
  46 static int log_level_details = 0;
  47 static int log_level_initialized = 0;
  48
  49 /* 1 if we use eliminitate identical delete/insert keys */
  50 /* eventually this the 0-case code will be removed */
  51 #define FLUSH2 1
  52
  53 void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
  54                                 zebra_rec_keys_t ins_keys,
  55                                 zint ins_rank,
  56                                 zebra_rec_keys_t del_keys,
  57                                 zint del_rank);
  58
  59 static void zebra_init_log_level(void)
  60 {
  61     if (!log_level_initialized)
  62     {
  63         log_level_initialized = 1;
  64
  65         log_level_extract = yaz_log_module_level("extract");
  66         log_level_details = yaz_log_module_level("indexdetails");
  67     }
  68 }
  69
  70 static void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
  71                                     int cmd, zebra_rec_keys_t skp);
  72 static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid);
  73 static void extract_token_add(RecWord *p);
  74
  75 static void check_log_limit(ZebraHandle zh)
  76 {
  77     if (zh->records_processed + zh->records_skipped == zh->m_file_verbose_limit)
  78     {
  79         yaz_log(YLOG_LOG, "More than %d file log entries. Omitting rest",
  80                 zh->m_file_verbose_limit);
  81     }
  82 }
  83
  84 static void logRecord(ZebraHandle zh)
  85 {
  86     check_log_limit(zh);
  87     ++zh->records_processed;
  88     if (!(zh->records_processed % 1000))
  89     {
  90         yaz_log(YLOG_LOG, "Records: "ZINT_FORMAT" i/u/d "
  91                 ZINT_FORMAT"/"ZINT_FORMAT"/"ZINT_FORMAT,
  92                 zh->records_processed, zh->records_inserted,
  93                 zh->records_updated, zh->records_deleted);
  94     }
  95 }
  96
  97 static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
  98 {
  99     ctrl->flagShowRecords = !zh->m_flag_rw;
 100 }
 101
 102
 103 static void extract_add_index_string(RecWord *p,
 104                                       zinfo_index_category_t cat,
 105                                       const char *str, int length);
 106
 107 static void extract_set_store_data_prepare(struct recExtractCtrl *p);
 108
 109 static void extract_init(struct recExtractCtrl *p, RecWord *w)
 110 {
 111     w->seqno = 1;
 112     w->index_name = "any";
 113     w->index_type = "w";
 114     w->extractCtrl = p;
 115     w->record_id = 0;
 116     w->section_id = 0;
 117     w->segment = 0;
 118 }
 119
 120 struct snip_rec_info {
 121     ZebraHandle zh;
 122     zebra_snippets *snippets;
 123 };
 124
 125
 126 static void snippet_add_complete_field(RecWord *p, int ord,
 127                                        zebra_map_t zm)
 128 {
 129     struct snip_rec_info *h = p->extractCtrl->handle;
 130
 131     const char *b = p->term_buf;
 132     char buf[IT_MAX_WORD+1];
 133     const char **map = 0;
 134     int i = 0, remain = p->term_len;
 135     const char *start = b;
 136     const char *last = 0;
 137
 138     if (remain > 0)
 139         map = zebra_maps_input(zm, &b, remain, 1);
 140
 141     while (remain > 0 && i < IT_MAX_WORD)
 142     {
 143         while (map && *map && **map == *CHR_SPACE)
 144         {
 145             remain = p->term_len - (b - p->term_buf);
 146
 147             if (i == 0)
 148                 start = b;  /* set to first non-ws area */
 149             if (remain > 0)
 150             {
 151                 int first = i ? 0 : 1;  /* first position */
 152
 153                 map = zebra_maps_input(zm, &b, remain, first);
 154             }
 155             else
 156                 map = 0;
 157         }
 158         if (!map)
 159             break;
 160
 161         if (i && i < IT_MAX_WORD)
 162             buf[i++] = *CHR_SPACE;
 163         while (map && *map && **map != *CHR_SPACE)
 164         {
 165             const char *cp = *map;
 166
 167             if (**map == *CHR_CUT)
 168             {
 169                 i = 0;
 170             }
 171             else
 172             {
 173                 if (i >= IT_MAX_WORD)
 174                     break;
 175                 while (i < IT_MAX_WORD && *cp)
 176                     buf[i++] = *(cp++);
 177             }
 178             last = b;
 179             remain = p->term_len  - (b - p->term_buf);
 180             if (remain > 0)
 181             {
 182                 map = zebra_maps_input(zm, &b, remain, 0);
 183             }
 184             else
 185                 map = 0;
 186         }
 187     }
 188     if (!i)
 189         return;
 190     if (last && start != last && zebra_maps_is_index(zm))
 191         zebra_snippets_appendn(h->snippets, p->seqno, 0, ord,
 192                                start, last - start);
 193 }
 194
 195 static void snippet_add_incomplete_field(RecWord *p, int ord, zebra_map_t zm)
 196 {
 197     struct snip_rec_info *h = p->extractCtrl->handle;
 198     const char *b = p->term_buf;
 199     int remain = p->term_len;
 200     int first = 1;
 201     const char **map = 0;
 202     const char *start = b;
 203     const char *last = b;
 204
 205     if (remain > 0)
 206         map = zebra_maps_input(zm, &b, remain, 0);
 207
 208     while (map)
 209     {
 210         char buf[IT_MAX_WORD+1];
 211         int i, remain;
 212
 213         /* Skip spaces */
 214         while (map && *map && **map == *CHR_SPACE)
 215         {
 216             remain = p->term_len - (b - p->term_buf);
 217             last = b;
 218             if (remain > 0)
 219                 map = zebra_maps_input(zm, &b, remain, 0);
 220             else
 221                 map = 0;
 222         }
 223         if (!map)
 224             break;
 225         if (start != last && zebra_maps_is_index(zm))
 226         {
 227             zebra_snippets_appendn(h->snippets, p->seqno, 1, ord,
 228                                    start, last - start);
 229
 230         }
 231         start = last;
 232
 233         i = 0;
 234         while (map && *map && **map != *CHR_SPACE)
 235         {
 236             const char *cp = *map;
 237
 238             while (i < IT_MAX_WORD && *cp)
 239                 buf[i++] = *(cp++);
 240             remain = p->term_len - (b - p->term_buf);
 241             last = b;
 242             if (remain > 0)
 243                 map = zebra_maps_input(zm, &b, remain, 0);
 244             else
 245                 map = 0;
 246         }
 247         if (!i)
 248             return;
 249
 250         if (first)
 251         {
 252             first = 0;
 253             if (zebra_maps_is_first_in_field(zm))
 254             {
 255                 /* first in field marker */
 256                 p->seqno++;
 257             }
 258         }
 259         if (start != last && zebra_maps_is_index(zm))
 260             zebra_snippets_appendn(h->snippets, p->seqno, 0, ord,
 261                                    start, last - start);
 262         start = last;
 263         p->seqno++;
 264     }
 265
 266 }
 267
 268 static void snippet_add_icu(RecWord *p, int ord, zebra_map_t zm)
 269 {
 270     struct snip_rec_info *h = p->extractCtrl->handle;
 271
 272     const char *res_buf = 0;
 273     size_t res_len = 0;
 274
 275     const char *display_buf = 0;
 276     size_t display_len = 0;
 277
 278     zebra_map_tokenize_start(zm, p->term_buf, p->term_len);
 279     while (zebra_map_tokenize_next(zm, &res_buf, &res_len,
 280                                    &display_buf, &display_len))
 281     {
 282         if (zebra_maps_is_index(zm))
 283             zebra_snippets_appendn(h->snippets, p->seqno, 0, ord,
 284                                    display_buf, display_len);
 285         p->seqno++;
 286     }
 287 }
 288
 289 static void snippet_token_add(RecWord *p)
 290 {
 291     struct snip_rec_info *h = p->extractCtrl->handle;
 292     ZebraHandle zh = h->zh;
 293     zebra_map_t zm = zebra_map_get(zh->reg->zebra_maps, p->index_type);
 294
 295     if (zm)
 296     {
 297         ZebraExplainInfo zei = zh->reg->zei;
 298         int ch = zebraExplain_lookup_attr_str(
 299             zei, zinfo_index_category_index, p->index_type, p->index_name);
 300
 301         if (zebra_maps_is_icu(zm))
 302             snippet_add_icu(p, ch, zm);
 303         else
 304         {
 305             if (zebra_maps_is_complete(zm))
 306                 snippet_add_complete_field(p, ch, zm);
 307             else
 308                 snippet_add_incomplete_field(p, ch, zm);
 309         }
 310     }
 311 }
 312
 313 static void snippet_schema_add(
 314     struct recExtractCtrl *p, Odr_oid *oid)
 315 {
 316
 317 }
 318
 319 void extract_snippet(ZebraHandle zh, zebra_snippets *sn,
 320                      struct ZebraRecStream *stream,
 321                      RecType rt, void *recTypeClientData)
 322 {
 323     struct recExtractCtrl extractCtrl;
 324     struct snip_rec_info info;
 325     int r;
 326
 327     extractCtrl.stream = stream;
 328     extractCtrl.first_record = 1;
 329     extractCtrl.init = extract_init;
 330     extractCtrl.tokenAdd = snippet_token_add;
 331     extractCtrl.schemaAdd = snippet_schema_add;
 332     assert(zh->reg);
 333     assert(zh->reg->dh);
 334
 335     extractCtrl.dh = zh->reg->dh;
 336
 337     info.zh = zh;
 338     info.snippets = sn;
 339     extractCtrl.handle = &info;
 340     extractCtrl.match_criteria[0] = '\0';
 341     extractCtrl.staticrank = 0;
 342     extractCtrl.action = action_insert;
 343
 344     init_extractCtrl(zh, &extractCtrl);
 345
 346     extractCtrl.setStoreData = 0;
 347
 348     r = (*rt->extract)(recTypeClientData, &extractCtrl);
 349
 350 }
 351
 352 static void searchRecordKey(ZebraHandle zh,
 353                             zebra_rec_keys_t reckeys,
 354                             const char *index_name,
 355                             const char **ws, int ws_length)
 356 {
 357     int i;
 358     int ch = -1;
 359     zinfo_index_category_t cat = zinfo_index_category_index;
 360
 361     for (i = 0; i<ws_length; i++)
 362         ws[i] = NULL;
 363
 364     if (ch < 0)
 365         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, "0", index_name);
 366     if (ch < 0)
 367         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, "p", index_name);
 368     if (ch < 0)
 369         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, "w", index_name);
 370
 371     if (ch < 0)
 372         return ;
 373
 374     if (zebra_rec_keys_rewind(reckeys))
 375     {
 376         zint startSeq = -1;
 377         const char *str;
 378         size_t slen;
 379         struct it_key key;
 380         zint seqno;
 381         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
 382         {
 383             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
 384
 385             seqno = key.mem[key.len-1];
 386
 387             if (key.mem[0] == ch)
 388             {
 389                 zint woff;
 390
 391                 if (startSeq == -1)
 392                     startSeq = seqno;
 393                 woff = seqno - startSeq;
 394                 if (woff >= 0 && woff < ws_length)
 395                     ws[woff] = str;
 396             }
 397         }
 398     }
 399 }
 400
 401 #define FILE_MATCH_BLANK "\t "
 402
 403 static char *get_match_from_spec(ZebraHandle zh,
 404                           zebra_rec_keys_t reckeys,
 405                           const char *fname, const char *spec)
 406 {
 407     static char dstBuf[2048];      /* static here ??? */
 408     char *dst = dstBuf;
 409     const char *s = spec;
 410
 411     while (1)
 412     {
 413         for (; *s && strchr(FILE_MATCH_BLANK, *s); s++)
 414             ;
 415         if (!*s)
 416             break;
 417         if (*s == '(')
 418         {
 419             const char *ws[32];
 420             char attset_str[64], attname_str[64];
 421             int i;
 422             int first = 1;
 423
 424             for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
 425                 ;
 426             for (i = 0; *s && *s != ',' && *s != ')' &&
 427                      !strchr(FILE_MATCH_BLANK, *s); s++)
 428                 if (i+1 < sizeof(attset_str))
 429                     attset_str[i++] = *s;
 430             attset_str[i] = '\0';
 431
 432             for (; strchr(FILE_MATCH_BLANK, *s); s++)
 433                 ;
 434             if (*s != ',')
 435                 strcpy(attname_str, attset_str);
 436             else
 437             {
 438                 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
 439                     ;
 440                 for (i = 0; *s && *s != ')' &&
 441                          !strchr(FILE_MATCH_BLANK, *s); s++)
 442                     if (i+1 < sizeof(attname_str))
 443                         attname_str[i++] = *s;
 444                 attname_str[i] = '\0';
 445             }
 446
 447             searchRecordKey(zh, reckeys, attname_str, ws, 32);
 448
 449             if (*s != ')')
 450             {
 451                 yaz_log(YLOG_WARN, "Missing ) in match criteria %s in group %s",
 452                       spec, zh->m_group ? zh->m_group : "none");
 453                 return NULL;
 454             }
 455             s++;
 456
 457             for (i = 0; i<32; i++)
 458                 if (ws[i])
 459                 {
 460                     if (first)
 461                     {
 462                         *dst++ = ' ';
 463                         first = 0;
 464                     }
 465                     strcpy(dst, ws[i]);
 466                     dst += strlen(ws[i]);
 467                 }
 468             if (first)
 469             {
 470                 yaz_log(YLOG_WARN, "Record didn't contain match"
 471                       " fields in (%s,%s)", attset_str, attname_str);
 472                 return NULL;
 473             }
 474         }
 475         else if (*s == '$')
 476         {
 477             int spec_len;
 478             char special[64];
 479             const char *spec_src = NULL;
 480             const char *s1 = ++s;
 481             while (*s1 && !strchr(FILE_MATCH_BLANK, *s1))
 482                 s1++;
 483
 484             spec_len = s1 - s;
 485             if (spec_len > sizeof(special)-1)
 486                 spec_len = sizeof(special)-1;
 487             memcpy(special, s, spec_len);
 488             special[spec_len] = '\0';
 489             s = s1;
 490
 491             if (!strcmp(special, "group"))
 492                 spec_src = zh->m_group;
 493             else if (!strcmp(special, "database"))
 494                 spec_src = zh->basenames[0];
 495             else if (!strcmp(special, "filename")) {
 496                 spec_src = fname;
 497             }
 498             else if (!strcmp(special, "type"))
 499                 spec_src = zh->m_record_type;
 500             else
 501                 spec_src = NULL;
 502             if (spec_src)
 503             {
 504                 strcpy(dst, spec_src);
 505                 dst += strlen(spec_src);
 506             }
 507         }
 508         else if (*s == '\"' || *s == '\'')
 509         {
 510             int stopMarker = *s++;
 511             char tmpString[64];
 512             int i = 0;
 513
 514             while (*s && *s != stopMarker)
 515             {
 516                 if (i+1 < sizeof(tmpString))
 517                     tmpString[i++] = *s++;
 518             }
 519             if (*s)
 520                 s++;
 521             tmpString[i] = '\0';
 522             strcpy(dst, tmpString);
 523             dst += strlen(tmpString);
 524         }
 525         else
 526         {
 527             yaz_log(YLOG_WARN, "Syntax error in match criteria %s in group %s",
 528                   spec, zh->m_group ? zh->m_group : "none");
 529             return NULL;
 530         }
 531         *dst++ = 1;
 532     }
 533     if (dst == dstBuf)
 534     {
 535         yaz_log(YLOG_WARN, "No match criteria for record %s in group %s",
 536               fname, zh->m_group ? zh->m_group : "none");
 537         return NULL;
 538     }
 539     *dst = '\0';
 540     return dstBuf;
 541 }
 542
 543 struct recordLogInfo {
 544     const char *fname;
 545     int recordOffset;
 546     struct recordGroup *rGroup;
 547 };
 548
 549 /** \brief add the always-matches index entry and map to real record ID
 550     \param ctrl record control
 551     \param record_id custom record ID
 552     \param sysno system record ID
 553
 554     This function serves two purposes.. It adds the always matches
 555     entry and makes a pointer from the custom record ID (if defined)
 556     back to the system record ID (sysno)
 557     See zebra_recid_to_sysno .
 558   */
 559 static void all_matches_add(struct recExtractCtrl *ctrl, zint record_id,
 560                             zint sysno)
 561 {
 562     RecWord word;
 563     extract_init(ctrl, &word);
 564     word.record_id = record_id;
 565     /* we use the seqno as placeholder for a way to get back to
 566        record database from _ALLRECORDS.. This is used if a custom
 567        RECORD was defined */
 568     word.seqno = sysno;
 569     word.index_name = "_ALLRECORDS";
 570     word.index_type = "w";
 571
 572     extract_add_index_string(&word, zinfo_index_category_alwaysmatches,
 573                               "", 0);
 574 }
 575
 576 ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
 577                                        struct ZebraRecStream *stream,
 578                                        enum zebra_recctrl_action_t action,
 579                                        int test_mode,
 580                                        const char *recordType,
 581                                        zint *sysno,
 582                                        const char *match_criteria,
 583                                        const char *fname,
 584                                        RecType recType,
 585                                        void *recTypeClientData);
 586
 587
 588 ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
 589                              enum zebra_recctrl_action_t action)
 590 {
 591     ZEBRA_RES r = ZEBRA_OK;
 592     int i, fd;
 593     char gprefix[128];
 594     char ext[128];
 595     char ext_res[128];
 596     struct file_read_info *fi = 0;
 597     const char *original_record_type = 0;
 598     RecType recType;
 599     void *recTypeClientData;
 600     struct ZebraRecStream stream, *streamp;
 601
 602     zebra_init_log_level();
 603
 604     if (!zh->m_group || !*zh->m_group)
 605         *gprefix = '\0';
 606     else
 607         sprintf(gprefix, "%s.", zh->m_group);
 608
 609     yaz_log(log_level_extract, "zebra_extract_file %s", fname);
 610
 611     /* determine file extension */
 612     *ext = '\0';
 613     for (i = strlen(fname); --i >= 0; )
 614         if (fname[i] == '/')
 615             break;
 616         else if (fname[i] == '.')
 617         {
 618             strcpy(ext, fname+i+1);
 619             break;
 620         }
 621     /* determine file type - depending on extension */
 622     original_record_type = zh->m_record_type;
 623     if (!zh->m_record_type)
 624     {
 625         sprintf(ext_res, "%srecordType.%s", gprefix, ext);
 626         zh->m_record_type = res_get(zh->res, ext_res);
 627     }
 628     if (!zh->m_record_type)
 629     {
 630         check_log_limit(zh);
 631         if (zh->records_processed + zh->records_skipped
 632             < zh->m_file_verbose_limit)
 633             yaz_log(YLOG_LOG, "? %s", fname);
 634         zh->records_skipped++;
 635         return 0;
 636     }
 637     /* determine match criteria */
 638     if (!zh->m_record_id)
 639     {
 640         sprintf(ext_res, "%srecordId.%s", gprefix, ext);
 641         zh->m_record_id = res_get(zh->res, ext_res);
 642     }
 643
 644     if (!(recType =
 645           recType_byName(zh->reg->recTypes, zh->res, zh->m_record_type,
 646                           &recTypeClientData)))
 647     {
 648         yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
 649         return ZEBRA_FAIL;
 650     }
 651
 652     switch(recType->version)
 653     {
 654     case 0:
 655         break;
 656     default:
 657         yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
 658     }
 659     if (sysno && (action == action_delete || action == action_a_delete))
 660     {
 661         streamp = 0;
 662         fi = 0;
 663     }
 664     else
 665     {
 666         char full_rep[1024];
 667
 668         if (zh->path_reg && !yaz_is_abspath(fname))
 669         {
 670             strcpy(full_rep, zh->path_reg);
 671             strcat(full_rep, "/");
 672             strcat(full_rep, fname);
 673         }
 674         else
 675             strcpy(full_rep, fname);
 676
 677         if ((fd = open(full_rep, O_BINARY|O_RDONLY)) == -1)
 678         {
 679             yaz_log(YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
 680             zh->m_record_type = original_record_type;
 681             return ZEBRA_FAIL;
 682         }
 683         streamp = &stream;
 684         zebra_create_stream_fd(streamp, fd, 0);
 685     }
 686     r = zebra_extract_records_stream(zh, streamp,
 687                                      action,
 688                                      0, /* tst_mode */
 689                                      zh->m_record_type,
 690                                      sysno,
 691                                      0, /*match_criteria */
 692                                      fname,
 693                                      recType, recTypeClientData);
 694     if (streamp)
 695         stream.destroy(streamp);
 696     zh->m_record_type = original_record_type;
 697     return r;
 698 }
 699
 700 /*
 701   If sysno is provided, then it's used to identify the reocord.
 702   If not, and match_criteria is provided, then sysno is guessed
 703   If not, and a record is provided, then sysno is got from there
 704
 705  */
 706
 707 ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh,
 708                                       const char *buf, size_t buf_size,
 709                                       enum zebra_recctrl_action_t action,
 710                                       int test_mode,
 711                                       const char *recordType,
 712                                       zint *sysno,
 713                                       const char *match_criteria,
 714                                       const char *fname)
 715 {
 716     struct ZebraRecStream stream;
 717     ZEBRA_RES res;
 718     void *clientData;
 719     RecType recType = 0;
 720
 721     if (recordType && *recordType)
 722     {
 723         yaz_log(log_level_extract,
 724                 "Record type explicitly specified: %s", recordType);
 725         recType = recType_byName(zh->reg->recTypes, zh->res, recordType,
 726                                   &clientData);
 727     }
 728     else
 729     {
 730         if (!(zh->m_record_type))
 731         {
 732             yaz_log(YLOG_WARN, "No such record type defined");
 733             return ZEBRA_FAIL;
 734         }
 735         yaz_log(log_level_extract, "Get record type from rgroup: %s",
 736                 zh->m_record_type);
 737         recType = recType_byName(zh->reg->recTypes, zh->res,
 738                                   zh->m_record_type, &clientData);
 739         recordType = zh->m_record_type;
 740     }
 741
 742     if (!recType)
 743     {
 744         yaz_log(YLOG_WARN, "No such record type: %s", recordType);
 745         return ZEBRA_FAIL;
 746     }
 747
 748     zebra_create_stream_mem(&stream, buf, buf_size);
 749
 750     res = zebra_extract_records_stream(zh, &stream,
 751                                        action,
 752                                        test_mode,
 753                                        recordType,
 754                                        sysno,
 755                                        match_criteria,
 756                                        fname,
 757                                        recType, clientData);
 758     stream.destroy(&stream);
 759     return res;
 760 }
 761
 762 ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
 763                                        struct ZebraRecStream *stream,
 764                                        enum zebra_recctrl_action_t action,
 765                                        int test_mode,
 766                                        const char *recordType,
 767                                        zint *sysno,
 768                                        const char *match_criteria,
 769                                        const char *fname,
 770                                        RecType recType,
 771                                        void *recTypeClientData)
 772 {
 773     ZEBRA_RES res = ZEBRA_OK;
 774     while (1)
 775     {
 776         int more = 0;
 777         res = zebra_extract_record_stream(zh, stream,
 778                                           action,
 779                                           test_mode,
 780                                           recordType,
 781                                           sysno,
 782                                           match_criteria,
 783                                           fname,
 784                                           recType, recTypeClientData, &more);
 785         if (!more)
 786         {
 787             res = ZEBRA_OK;
 788             break;
 789         }
 790         if (res != ZEBRA_OK)
 791             break;
 792         if (sysno)
 793             break;
 794     }
 795     return res;
 796 }
 797
 798
 799 static WRBUF wrbuf_hex_str(const char *cstr)
 800 {
 801     size_t i;
 802     WRBUF w = wrbuf_alloc();
 803     for (i = 0; cstr[i]; i++)
 804     {
 805         if (cstr[i] < ' ' || cstr[i] > 126)
 806             wrbuf_printf(w, "\\%02X", cstr[i] & 0xff);
 807         else
 808             wrbuf_putc(w, cstr[i]);
 809     }
 810     return w;
 811 }
 812
 813 ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
 814                                       struct ZebraRecStream *stream,
 815                                       enum zebra_recctrl_action_t action,
 816                                       int test_mode,
 817                                       const char *recordType,
 818                                       zint *sysno,
 819                                       const char *match_criteria,
 820                                       const char *fname,
 821                                       RecType recType,
 822                                       void *recTypeClientData,
 823                                       int *more)
 824
 825 {
 826     zint sysno0 = 0;
 827     RecordAttr *recordAttr;
 828     struct recExtractCtrl extractCtrl;
 829     int r;
 830     const char *matchStr = 0;
 831     Record rec;
 832     off_t start_offset = 0, end_offset = 0;
 833     const char *pr_fname = fname;  /* filename to print .. */
 834     int show_progress = zh->records_processed + zh->records_skipped
 835         < zh->m_file_verbose_limit ? 1:0;
 836
 837     zebra_init_log_level();
 838
 839     if (!pr_fname)
 840         pr_fname = "<no file>";  /* make it printable if file is omitted */
 841
 842     zebra_rec_keys_reset(zh->reg->keys);
 843     zebra_rec_keys_reset(zh->reg->sortKeys);
 844
 845     if (zebraExplain_curDatabase(zh->reg->zei, zh->basenames[0]))
 846     {
 847         if (zebraExplain_newDatabase(zh->reg->zei, zh->basenames[0],
 848                                       zh->m_explain_database))
 849             return ZEBRA_FAIL;
 850     }
 851
 852     if (stream)
 853     {
 854         off_t null_offset = 0;
 855         extractCtrl.stream = stream;
 856
 857         start_offset = stream->tellf(stream);
 858
 859         extractCtrl.first_record = start_offset ? 0 : 1;
 860
 861         stream->endf(stream, &null_offset);;
 862
 863         extractCtrl.init = extract_init;
 864         extractCtrl.tokenAdd = extract_token_add;
 865         extractCtrl.schemaAdd = extract_schema_add;
 866         extractCtrl.dh = zh->reg->dh;
 867         extractCtrl.handle = zh;
 868         extractCtrl.match_criteria[0] = '\0';
 869         extractCtrl.staticrank = 0;
 870         extractCtrl.action = action;
 871
 872         init_extractCtrl(zh, &extractCtrl);
 873
 874         extract_set_store_data_prepare(&extractCtrl);
 875
 876         r = (*recType->extract)(recTypeClientData, &extractCtrl);
 877
 878         if (action == action_update)
 879         {
 880             action = extractCtrl.action;
 881         }
 882
 883         switch (r)
 884         {
 885         case RECCTRL_EXTRACT_EOF:
 886             return ZEBRA_FAIL;
 887         case RECCTRL_EXTRACT_ERROR_GENERIC:
 888             /* error occured during extraction ... */
 889             yaz_log(YLOG_WARN, "extract error: generic");
 890             return ZEBRA_FAIL;
 891         case RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER:
 892             /* error occured during extraction ... */
 893             yaz_log(YLOG_WARN, "extract error: no such filter");
 894             return ZEBRA_FAIL;
 895         case RECCTRL_EXTRACT_SKIP:
 896             if (show_progress)
 897                 yaz_log(YLOG_LOG, "skip %s %s " ZINT_FORMAT,
 898                          recordType, pr_fname, (zint) start_offset);
 899             *more = 1;
 900
 901             end_offset = stream->endf(stream, 0);
 902             if (end_offset)
 903                 stream->seekf(stream, end_offset);
 904
 905             return ZEBRA_OK;
 906         case RECCTRL_EXTRACT_OK:
 907             break;
 908         default:
 909             yaz_log(YLOG_WARN, "extract error: unknown error: %d", r);
 910             return ZEBRA_FAIL;
 911         }
 912         end_offset = stream->endf(stream, 0);
 913         if (end_offset)
 914             stream->seekf(stream, end_offset);
 915         else
 916             end_offset = stream->tellf(stream);
 917
 918         if (extractCtrl.match_criteria[0])
 919             match_criteria = extractCtrl.match_criteria;
 920     }
 921
 922     *more = 1;
 923     if (!sysno)
 924     {
 925         sysno = &sysno0;
 926
 927         if (match_criteria && *match_criteria) {
 928             matchStr = match_criteria;
 929         } else {
 930             if (zh->m_record_id && *zh->m_record_id) {
 931                 matchStr = get_match_from_spec(zh, zh->reg->keys, pr_fname,
 932                                                zh->m_record_id);
 933                 if (!matchStr)
 934                 {
 935                     yaz_log(YLOG_LOG, "error %s %s " ZINT_FORMAT, recordType,
 936                              pr_fname, (zint) start_offset);
 937                     return ZEBRA_FAIL;
 938                 }
 939             }
 940         }
 941         if (matchStr)
 942         {
 943             int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
 944             char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
 945                                           matchStr);
 946
 947
 948             if (log_level_extract)
 949             {
 950                 WRBUF w = wrbuf_hex_str(matchStr);
 951                 yaz_log(log_level_extract, "matchStr: %s", wrbuf_cstr(w));
 952                 wrbuf_destroy(w);
 953             }
 954             if (rinfo)
 955             {
 956                 assert(*rinfo == sizeof(*sysno));
 957                 memcpy(sysno, rinfo+1, sizeof(*sysno));
 958             }
 959        }
 960     }
 961
 962     if (zebra_rec_keys_empty(zh->reg->keys))
 963     {
 964         /* the extraction process returned no information - the record
 965            is probably empty - unless flagShowRecords is in use */
 966         if (test_mode)
 967             return ZEBRA_OK;
 968     }
 969
 970     if (! *sysno)
 971     {
 972         /* new record AKA does not exist already */
 973         if (action == action_delete)
 974         {
 975                 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
 976                         pr_fname, (zint) start_offset);
 977             yaz_log(YLOG_WARN, "cannot delete record above (seems new)");
 978             return ZEBRA_FAIL;
 979         }
 980         else if (action == action_a_delete)
 981         {
 982             if (show_progress)
 983                 yaz_log(YLOG_LOG, "adelete %s %s " ZINT_FORMAT, recordType,
 984                         pr_fname, (zint) start_offset);
 985             return ZEBRA_OK;
 986         }
 987         else if (action == action_replace)
 988         {
 989             yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
 990                          pr_fname, (zint) start_offset);
 991             yaz_log(YLOG_WARN, "cannot update record above (seems new)");
 992             return ZEBRA_FAIL;
 993         }
 994         if (show_progress)
 995             yaz_log(YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname,
 996                      (zint) start_offset);
 997         rec = rec_new(zh->reg->records);
 998
 999         *sysno = rec->sysno;
1000
1001
1002         if (stream)
1003         {
1004             all_matches_add(&extractCtrl,
1005                             zebra_rec_keys_get_custom_record_id(zh->reg->keys),
1006                             *sysno);
1007         }
1008
1009
1010         recordAttr = rec_init_attr(zh->reg->zei, rec);
1011         if (extractCtrl.staticrank < 0)
1012         {
1013             yaz_log(YLOG_WARN, "Negative staticrank for record. Set to 0");
1014             extractCtrl.staticrank = 0;
1015         }
1016
1017         if (matchStr)
1018         {
1019             int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
1020             dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
1021                             sizeof(*sysno), sysno);
1022         }
1023
1024         extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
1025 #if FLUSH2
1026         extract_flush_record_keys2(zh, *sysno,
1027                                    zh->reg->keys, extractCtrl.staticrank,
1028                                    0, recordAttr->staticrank);
1029 #else
1030         extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys,
1031                                   extractCtrl.staticrank);
1032 #endif
1033         recordAttr->staticrank = extractCtrl.staticrank;
1034         zh->records_inserted++;
1035     }
1036     else
1037     {
1038         /* record already exists */
1039         zebra_rec_keys_t delkeys = zebra_rec_keys_open();
1040         zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
1041         if (action == action_insert)
1042         {
1043             yaz_log(YLOG_LOG, "skipped %s %s " ZINT_FORMAT,
1044                          recordType, pr_fname, (zint) start_offset);
1045             logRecord(zh);
1046             return ZEBRA_FAIL;
1047         }
1048
1049         rec = rec_get(zh->reg->records, *sysno);
1050         assert(rec);
1051
1052         if (stream)
1053         {
1054             all_matches_add(&extractCtrl,
1055                             zebra_rec_keys_get_custom_record_id(zh->reg->keys),
1056                             *sysno);
1057         }
1058
1059         recordAttr = rec_init_attr(zh->reg->zei, rec);
1060
1061         /* decrease total size */
1062         zebraExplain_recordBytesIncrement(zh->reg->zei,
1063                                            - recordAttr->recordSize);
1064
1065         zebra_rec_keys_set_buf(delkeys,
1066                                rec->info[recInfo_delKeys],
1067                                rec->size[recInfo_delKeys],
1068                                0);
1069         zebra_rec_keys_set_buf(sortKeys,
1070                                rec->info[recInfo_sortKeys],
1071                                rec->size[recInfo_sortKeys],
1072                                0);
1073
1074         extract_flush_sort_keys(zh, *sysno, 0, sortKeys);
1075 #if !FLUSH2
1076         extract_flush_record_keys(zh, *sysno, 0, delkeys,
1077                                   recordAttr->staticrank);
1078 #endif
1079         if (action == action_delete || action == action_a_delete)
1080         {
1081             /* record going to be deleted */
1082 #if FLUSH2
1083             extract_flush_record_keys2(zh, *sysno, 0, recordAttr->staticrank,
1084                                        delkeys, recordAttr->staticrank);
1085 #endif
1086             if (zebra_rec_keys_empty(delkeys))
1087             {
1088                 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
1089                         pr_fname, (zint) start_offset);
1090                 yaz_log(YLOG_WARN, "cannot delete file above, "
1091                         "storeKeys false (3)");
1092             }
1093             else
1094             {
1095                 if (show_progress)
1096                     yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
1097                             pr_fname, (zint) start_offset);
1098                 zh->records_deleted++;
1099                 if (matchStr)
1100                 {
1101                     int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
1102                     dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
1103                 }
1104                 rec_del(zh->reg->records, &rec);
1105             }
1106             zebra_rec_keys_close(delkeys);
1107             zebra_rec_keys_close(sortKeys);
1108             rec_free(&rec);
1109             logRecord(zh);
1110             return ZEBRA_OK;
1111         }
1112         else
1113         {   /* update or special_update */
1114             if (show_progress)
1115                 yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
1116                         pr_fname, (zint) start_offset);
1117             extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
1118
1119 #if FLUSH2
1120             extract_flush_record_keys2(zh, *sysno,
1121                                        zh->reg->keys, extractCtrl.staticrank,
1122                                        delkeys, recordAttr->staticrank);
1123 #else
1124             extract_flush_record_keys(zh, *sysno, 1,
1125                                       zh->reg->keys, extractCtrl.staticrank);
1126 #endif
1127             recordAttr->staticrank = extractCtrl.staticrank;
1128             zh->records_updated++;
1129         }
1130         zebra_rec_keys_close(delkeys);
1131         zebra_rec_keys_close(sortKeys);
1132     }
1133     /* update file type */
1134     xfree(rec->info[recInfo_fileType]);
1135     rec->info[recInfo_fileType] =
1136         rec_strdup(recordType, &rec->size[recInfo_fileType]);
1137
1138     /* update filename */
1139     xfree(rec->info[recInfo_filename]);
1140     rec->info[recInfo_filename] =
1141         rec_strdup(fname, &rec->size[recInfo_filename]);
1142
1143     /* update delete keys */
1144     xfree(rec->info[recInfo_delKeys]);
1145     if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
1146     {
1147         zebra_rec_keys_get_buf(zh->reg->keys,
1148                                &rec->info[recInfo_delKeys],
1149                                &rec->size[recInfo_delKeys]);
1150     }
1151     else
1152     {
1153         rec->info[recInfo_delKeys] = NULL;
1154         rec->size[recInfo_delKeys] = 0;
1155     }
1156     /* update sort keys */
1157     xfree(rec->info[recInfo_sortKeys]);
1158
1159     zebra_rec_keys_get_buf(zh->reg->sortKeys,
1160                            &rec->info[recInfo_sortKeys],
1161                            &rec->size[recInfo_sortKeys]);
1162
1163     if (stream)
1164     {
1165         recordAttr->recordSize = end_offset - start_offset;
1166         zebraExplain_recordBytesIncrement(zh->reg->zei,
1167                                           recordAttr->recordSize);
1168     }
1169
1170     /* set run-number for this record */
1171     recordAttr->runNumber =
1172         zebraExplain_runNumberIncrement(zh->reg->zei, 0);
1173
1174     /* update store data */
1175     xfree(rec->info[recInfo_storeData]);
1176
1177     /* update store data */
1178     if (zh->store_data_buf)
1179     {
1180         rec->size[recInfo_storeData] = zh->store_data_size;
1181         rec->info[recInfo_storeData] = zh->store_data_buf;
1182         zh->store_data_buf = 0;
1183         recordAttr->recordSize = zh->store_data_size;
1184     }
1185     else if (zh->m_store_data)
1186     {
1187         off_t cur_offset = stream->tellf(stream);
1188
1189         rec->size[recInfo_storeData] = recordAttr->recordSize;
1190         rec->info[recInfo_storeData] = (char *)
1191             xmalloc(recordAttr->recordSize);
1192         stream->seekf(stream, start_offset);
1193         stream->readf(stream, rec->info[recInfo_storeData],
1194                       recordAttr->recordSize);
1195         stream->seekf(stream, cur_offset);
1196     }
1197     else
1198     {
1199         rec->info[recInfo_storeData] = NULL;
1200         rec->size[recInfo_storeData] = 0;
1201     }
1202     /* update database name */
1203     xfree(rec->info[recInfo_databaseName]);
1204     rec->info[recInfo_databaseName] =
1205         rec_strdup(zh->basenames[0], &rec->size[recInfo_databaseName]);
1206
1207     /* update offset */
1208     recordAttr->recordOffset = start_offset;
1209
1210     /* commit this record */
1211     rec_put(zh->reg->records, &rec);
1212     logRecord(zh);
1213     return ZEBRA_OK;
1214 }
1215
1216 ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
1217 {
1218     ZebraHandle zh = (ZebraHandle) handle;
1219     struct recExtractCtrl extractCtrl;
1220
1221     if (zebraExplain_curDatabase(zh->reg->zei,
1222                                   rec->info[recInfo_databaseName]))
1223     {
1224         abort();
1225         if (zebraExplain_newDatabase(zh->reg->zei,
1226                                       rec->info[recInfo_databaseName], 0))
1227             abort();
1228     }
1229
1230     zebra_rec_keys_reset(zh->reg->keys);
1231     zebra_rec_keys_reset(zh->reg->sortKeys);
1232
1233     extractCtrl.init = extract_init;
1234     extractCtrl.tokenAdd = extract_token_add;
1235     extractCtrl.schemaAdd = extract_schema_add;
1236     extractCtrl.dh = zh->reg->dh;
1237
1238     init_extractCtrl(zh, &extractCtrl);
1239
1240     extractCtrl.flagShowRecords = 0;
1241     extractCtrl.match_criteria[0] = '\0';
1242     extractCtrl.staticrank = 0;
1243     extractCtrl.action = action_update;
1244
1245     extractCtrl.handle = handle;
1246     extractCtrl.first_record = 1;
1247
1248     extract_set_store_data_prepare(&extractCtrl);
1249
1250     if (n)
1251         grs_extract_tree(&extractCtrl, n);
1252
1253     if (rec->size[recInfo_delKeys])
1254     {
1255         zebra_rec_keys_t delkeys = zebra_rec_keys_open();
1256
1257         zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
1258
1259         zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
1260                                rec->size[recInfo_delKeys],
1261                                0);
1262 #if FLUSH2
1263         extract_flush_record_keys2(zh, rec->sysno,
1264                                    zh->reg->keys, 0, delkeys, 0);
1265 #else
1266         extract_flush_record_keys(zh, rec->sysno, 0, delkeys, 0);
1267         extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
1268 #endif
1269         zebra_rec_keys_close(delkeys);
1270
1271         zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
1272                                rec->size[recInfo_sortKeys],
1273                                0);
1274
1275         extract_flush_sort_keys(zh, rec->sysno, 0, sortkeys);
1276         zebra_rec_keys_close(sortkeys);
1277     }
1278     else
1279     {
1280 #if FLUSH2
1281         extract_flush_record_keys2(zh, rec->sysno, zh->reg->keys, 0, 0, 0);
1282 #else
1283         extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
1284 #endif
1285     }
1286     extract_flush_sort_keys(zh, rec->sysno, 1, zh->reg->sortKeys);
1287
1288     xfree(rec->info[recInfo_delKeys]);
1289     zebra_rec_keys_get_buf(zh->reg->keys,
1290                            &rec->info[recInfo_delKeys],
1291                            &rec->size[recInfo_delKeys]);
1292
1293     xfree(rec->info[recInfo_sortKeys]);
1294     zebra_rec_keys_get_buf(zh->reg->sortKeys,
1295                            &rec->info[recInfo_sortKeys],
1296                            &rec->size[recInfo_sortKeys]);
1297     return ZEBRA_OK;
1298 }
1299
1300 void extract_rec_keys_log(ZebraHandle zh, int is_insert,
1301                           zebra_rec_keys_t reckeys,
1302                           int level)
1303 {
1304     if (zebra_rec_keys_rewind(reckeys))
1305     {
1306         size_t slen;
1307         const char *str;
1308         struct it_key key;
1309         NMEM nmem = nmem_create();
1310
1311         while(zebra_rec_keys_read(reckeys, &str, &slen, &key))
1312         {
1313             char keystr[200]; /* room for zints to print */
1314             char *dst_term = 0;
1315             int ord = CAST_ZINT_TO_INT(key.mem[0]);
1316             const char *index_type;
1317             int i;
1318             const char *string_index;
1319
1320             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1321                                     0/* db */, &string_index);
1322             assert(index_type);
1323             zebra_term_untrans_iconv(zh, nmem, index_type,
1324                                      &dst_term, str);
1325             *keystr = '\0';
1326             for (i = 0; i<key.len; i++)
1327             {
1328                 sprintf(keystr + strlen(keystr), ZINT_FORMAT " ", key.mem[i]);
1329             }
1330
1331             if (*str < CHR_BASE_CHAR)
1332             {
1333                 int i;
1334                 char dst_buf[200]; /* room for special chars */
1335
1336                 strcpy(dst_buf , "?");
1337
1338                 if (!strcmp(str, ""))
1339                     strcpy(dst_buf, "alwaysmatches");
1340                 if (!strcmp(str, FIRST_IN_FIELD_STR))
1341                     strcpy(dst_buf, "firstinfield");
1342                 else if (!strcmp(str, CHR_UNKNOWN))
1343                     strcpy(dst_buf, "unknown");
1344                 else if (!strcmp(str, CHR_SPACE))
1345                     strcpy(dst_buf, "space");
1346
1347                 for (i = 0; i<slen; i++)
1348                 {
1349                     sprintf(dst_buf + strlen(dst_buf), " %d", str[i] & 0xff);
1350                 }
1351                 yaz_log(level, "%s%s %s %s", keystr, index_type,
1352                         string_index, dst_buf);
1353
1354             }
1355             else
1356                 yaz_log(level, "%s%s %s \"%s\"", keystr, index_type,
1357                         string_index, dst_term);
1358
1359             nmem_reset(nmem);
1360         }
1361         nmem_destroy(nmem);
1362     }
1363 }
1364
1365 void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
1366                              zebra_rec_keys_t reckeys)
1367 {
1368     ZebraExplainInfo zei = zh->reg->zei;
1369     struct ord_stat {
1370         int no;
1371         int ord;
1372         struct ord_stat *next;
1373     };
1374
1375     if (zebra_rec_keys_rewind(reckeys))
1376     {
1377         struct ord_stat *ord_list = 0;
1378         struct ord_stat *p;
1379         size_t slen;
1380         const char *str;
1381         struct it_key key_in;
1382         while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1383         {
1384             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1385
1386             for (p = ord_list; p ; p = p->next)
1387                 if (p->ord == ord)
1388                 {
1389                     p->no++;
1390                     break;
1391                 }
1392             if (!p)
1393             {
1394                 p = xmalloc(sizeof(*p));
1395                 p->no = 1;
1396                 p->ord = ord;
1397                 p->next = ord_list;
1398                 ord_list = p;
1399             }
1400         }
1401
1402         p = ord_list;
1403         while (p)
1404         {
1405             struct ord_stat *p1 = p;
1406
1407             if (is_insert)
1408                 zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1);
1409             else
1410                 zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1);
1411             p = p->next;
1412             xfree(p1);
1413         }
1414     }
1415 }
1416
1417 void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
1418                                 zebra_rec_keys_t ins_keys, zint ins_rank,
1419                                 zebra_rec_keys_t del_keys, zint del_rank)
1420 {
1421     ZebraExplainInfo zei = zh->reg->zei;
1422     int normal = 0;
1423     int optimized = 0;
1424
1425     if (!zh->reg->key_block)
1426     {
1427         int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8"));
1428         const char *key_tmp_dir = res_get_def(zh->res, "keyTmpDir", ".");
1429         int use_threads = atoi(res_get_def(zh->res, "threads", "1"));
1430         zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads);
1431     }
1432
1433     if (ins_keys)
1434     {
1435         extract_rec_keys_adjust(zh, 1, ins_keys);
1436         if (!del_keys)
1437             zebraExplain_recordCountIncrement(zei, 1);
1438         zebra_rec_keys_rewind(ins_keys);
1439     }
1440     if (del_keys)
1441     {
1442         extract_rec_keys_adjust(zh, 0, del_keys);
1443         if (!ins_keys)
1444             zebraExplain_recordCountIncrement(zei, -1);
1445         zebra_rec_keys_rewind(del_keys);
1446     }
1447
1448     while (1)
1449     {
1450         size_t del_slen;
1451         const char *del_str;
1452         struct it_key del_key_in;
1453         int del = 0;
1454
1455         size_t ins_slen;
1456         const char *ins_str;
1457         struct it_key ins_key_in;
1458         int ins = 0;
1459
1460         if (del_keys)
1461             del = zebra_rec_keys_read(del_keys, &del_str, &del_slen,
1462                                       &del_key_in);
1463         if (ins_keys)
1464             ins = zebra_rec_keys_read(ins_keys, &ins_str, &ins_slen,
1465                                       &ins_key_in);
1466
1467         if (del && ins && ins_rank == del_rank
1468             && !key_compare(&del_key_in, &ins_key_in)
1469             && ins_slen == del_slen && !memcmp(del_str, ins_str, del_slen))
1470         {
1471             optimized++;
1472             continue;
1473         }
1474         if (!del && !ins)
1475             break;
1476
1477         normal++;
1478         if (del)
1479             key_block_write(zh->reg->key_block, sysno,
1480                             &del_key_in, 0, del_str, del_slen,
1481                             del_rank, zh->m_staticrank);
1482         if (ins)
1483             key_block_write(zh->reg->key_block, sysno,
1484                             &ins_key_in, 1, ins_str, ins_slen,
1485                             ins_rank, zh->m_staticrank);
1486     }
1487     yaz_log(log_level_extract, "normal=%d optimized=%d", normal, optimized);
1488 }
1489
1490
1491 ZEBRA_RES zebra_rec_keys_to_snippets1(ZebraHandle zh,
1492                                      zebra_rec_keys_t reckeys,
1493                                      zebra_snippets *snippets)
1494 {
1495     NMEM nmem = nmem_create();
1496     if (zebra_rec_keys_rewind(reckeys))
1497     {
1498         const char *str;
1499         size_t slen;
1500         struct it_key key;
1501         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1502         {
1503             char *dst_term = 0;
1504             int ord;
1505             zint seqno;
1506             const char *index_type;
1507
1508             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1509             seqno = key.mem[key.len-1];
1510             ord = CAST_ZINT_TO_INT(key.mem[0]);
1511
1512             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1513                                     0/* db */, 0 /* string_index */);
1514             assert(index_type);
1515             zebra_term_untrans_iconv(zh, nmem, index_type,
1516                                      &dst_term, str);
1517             zebra_snippets_append(snippets, seqno, 0, ord, dst_term);
1518             nmem_reset(nmem);
1519         }
1520     }
1521     nmem_destroy(nmem);
1522     return ZEBRA_OK;
1523 }
1524
1525 void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
1526 {
1527     yaz_log(YLOG_LOG, "print_rec_keys");
1528     if (zebra_rec_keys_rewind(reckeys))
1529     {
1530         const char *str;
1531         size_t slen;
1532         struct it_key key;
1533         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1534         {
1535             char dst_buf[IT_MAX_WORD];
1536             zint seqno;
1537             const char *index_type;
1538             int ord = CAST_ZINT_TO_INT(key.mem[0]);
1539             const char *db = 0;
1540             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1541
1542             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0);
1543
1544             seqno = key.mem[key.len-1];
1545
1546             zebra_term_untrans(zh, index_type, dst_buf, str);
1547
1548             yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT
1549                     " term=%s", ord, seqno, dst_buf);
1550         }
1551     }
1552 }
1553
1554 static void extract_add_index_string(RecWord *p, zinfo_index_category_t cat,
1555                                      const char *str, int length)
1556 {
1557     struct it_key key;
1558     ZebraHandle zh = p->extractCtrl->handle;
1559     ZebraExplainInfo zei = zh->reg->zei;
1560     int ch, i;
1561
1562     ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1563     if (ch < 0)
1564         ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1565
1566     i = 0;
1567     key.mem[i++] = ch;
1568     key.mem[i++] = p->record_id;
1569     key.mem[i++] = p->section_id;
1570
1571     if (zh->m_segment_indexing)
1572         key.mem[i++] = p->segment;
1573     key.mem[i++] = p->seqno;
1574     key.len = i;
1575
1576     zebra_rec_keys_write(zh->reg->keys, str, length, &key);
1577 }
1578
1579 static void extract_add_sort_string(RecWord *p, const char *str, int length)
1580 {
1581     struct it_key key;
1582     ZebraHandle zh = p->extractCtrl->handle;
1583     ZebraExplainInfo zei = zh->reg->zei;
1584     int ch;
1585     zinfo_index_category_t cat = zinfo_index_category_sort;
1586
1587     ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1588     if (ch < 0)
1589         ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1590     key.len = 2;
1591     key.mem[0] = ch;
1592     key.mem[1] = p->record_id;
1593
1594     zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
1595 }
1596
1597 static void extract_add_staticrank_string(RecWord *p,
1598                                           const char *str, int length)
1599 {
1600     char valz[40];
1601     struct recExtractCtrl *ctrl = p->extractCtrl;
1602
1603     if (length > sizeof(valz)-1)
1604         length = sizeof(valz)-1;
1605
1606     memcpy(valz, str, length);
1607     valz[length] = '\0';
1608     ctrl->staticrank = atozint(valz);
1609 }
1610
1611 static void extract_add_string(RecWord *p, zebra_map_t zm,
1612                                const char *string, int length)
1613 {
1614     assert(length > 0);
1615
1616     if (!p->index_name)
1617         return;
1618     if (log_level_details)
1619     {
1620
1621         WRBUF w = wrbuf_alloc();
1622
1623         wrbuf_write_escaped(w, string, length);
1624         yaz_log(log_level_details, "extract_add_string: %s", wrbuf_cstr(w));
1625         wrbuf_destroy(w);
1626     }
1627     if (zebra_maps_is_index(zm))
1628     {
1629         extract_add_index_string(p, zinfo_index_category_index,
1630                                  string, length);
1631         if (zebra_maps_is_alwaysmatches(zm))
1632         {
1633             RecWord word;
1634             memcpy(&word, p, sizeof(word));
1635
1636             word.seqno = 1;
1637             extract_add_index_string(
1638                 &word, zinfo_index_category_alwaysmatches, "", 0);
1639         }
1640     }
1641     else if (zebra_maps_is_sort(zm))
1642     {
1643         extract_add_sort_string(p, string, length);
1644     }
1645     else if (zebra_maps_is_staticrank(zm))
1646     {
1647         extract_add_staticrank_string(p, string, length);
1648     }
1649 }
1650
1651 static void extract_add_incomplete_field(RecWord *p, zebra_map_t zm)
1652 {
1653     const char *b = p->term_buf;
1654     int remain = p->term_len;
1655     int first = 1;
1656     const char **map = 0;
1657
1658     if (remain > 0)
1659         map = zebra_maps_input(zm, &b, remain, 0);
1660
1661     while (map)
1662     {
1663         char buf[IT_MAX_WORD+1];
1664         int i, remain;
1665
1666         /* Skip spaces */
1667         while (map && *map && **map == *CHR_SPACE)
1668         {
1669             remain = p->term_len - (b - p->term_buf);
1670             if (remain > 0)
1671                 map = zebra_maps_input(zm, &b, remain, 0);
1672             else
1673                 map = 0;
1674         }
1675         if (!map)
1676             break;
1677         i = 0;
1678         while (map && *map && **map != *CHR_SPACE)
1679         {
1680             const char *cp = *map;
1681
1682             while (i < IT_MAX_WORD && *cp)
1683                 buf[i++] = *(cp++);
1684             remain = p->term_len - (b - p->term_buf);
1685             if (remain > 0)
1686                 map = zebra_maps_input(zm, &b, remain, 0);
1687             else
1688                 map = 0;
1689         }
1690         if (!i)
1691             return;
1692
1693         if (first)
1694         {
1695             first = 0;
1696             if (zebra_maps_is_first_in_field(zm))
1697             {
1698                 /* first in field marker */
1699                 extract_add_string(p, zm, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN);
1700                 p->seqno++;
1701             }
1702         }
1703         extract_add_string(p, zm, buf, i);
1704         p->seqno++;
1705     }
1706 }
1707
1708 static void extract_add_complete_field(RecWord *p, zebra_map_t zm)
1709 {
1710     const char *b = p->term_buf;
1711     char buf[IT_MAX_WORD+1];
1712     const char **map = 0;
1713     int i = 0, remain = p->term_len;
1714
1715     if (remain > 0)
1716         map = zebra_maps_input(zm, &b, remain, 1);
1717
1718     while (remain > 0 && i < IT_MAX_WORD)
1719     {
1720         while (map && *map && **map == *CHR_SPACE)
1721         {
1722             remain = p->term_len - (b - p->term_buf);
1723
1724             if (remain > 0)
1725             {
1726                 int first = i ? 0 : 1;  /* first position */
1727                 map = zebra_maps_input(zm, &b, remain, first);
1728             }
1729             else
1730                 map = 0;
1731         }
1732         if (!map)
1733             break;
1734
1735         if (i && i < IT_MAX_WORD)
1736             buf[i++] = *CHR_SPACE;
1737         while (map && *map && **map != *CHR_SPACE)
1738         {
1739             const char *cp = *map;
1740
1741             if (**map == *CHR_CUT)
1742             {
1743                 i = 0;
1744             }
1745             else
1746             {
1747                 if (i >= IT_MAX_WORD)
1748                     break;
1749                 while (i < IT_MAX_WORD && *cp)
1750                     buf[i++] = *(cp++);
1751             }
1752             remain = p->term_len  - (b - p->term_buf);
1753             if (remain > 0)
1754             {
1755                 map = zebra_maps_input(zm, &b, remain, 0);
1756             }
1757             else
1758                 map = 0;
1759         }
1760     }
1761     if (!i)
1762         return;
1763     extract_add_string(p, zm, buf, i);
1764 }
1765
1766 static void extract_add_icu(RecWord *p, zebra_map_t zm)
1767 {
1768     const char *res_buf = 0;
1769     size_t res_len = 0;
1770
1771     zebra_map_tokenize_start(zm, p->term_buf, p->term_len);
1772     while (zebra_map_tokenize_next(zm, &res_buf, &res_len, 0, 0))
1773     {
1774         extract_add_string(p, zm, res_buf, res_len);
1775         p->seqno++;
1776     }
1777 }
1778
1779
1780 /** \brief top-level indexing handler for recctrl system
1781     \param p token data to be indexed
1782
1783     Call sequence:
1784     extract_token_add
1785     extract_add_{in}_complete / extract_add_icu
1786     extract_add_string
1787
1788     extract_add_index_string
1789     or
1790     extract_add_sort_string
1791     or
1792     extract_add_staticrank_string
1793
1794 */
1795 static void extract_token_add(RecWord *p)
1796 {
1797     ZebraHandle zh = p->extractCtrl->handle;
1798     zebra_map_t zm = zebra_map_get_or_add(zh->reg->zebra_maps, p->index_type);
1799     WRBUF wrbuf;
1800
1801     if (log_level_details)
1802     {
1803         yaz_log(log_level_details, "extract_token_add "
1804                 "type=%s index=%s seqno=" ZINT_FORMAT " s=%.*s",
1805                 p->index_type, p->index_name,
1806                 p->seqno, p->term_len, p->term_buf);
1807     }
1808     if ((wrbuf = zebra_replace(zm, 0, p->term_buf, p->term_len)))
1809     {
1810         p->term_buf = wrbuf_buf(wrbuf);
1811         p->term_len = wrbuf_len(wrbuf);
1812     }
1813     if (zebra_maps_is_icu(zm))
1814     {
1815         extract_add_icu(p, zm);
1816     }
1817     else
1818     {
1819         if (zebra_maps_is_complete(zm))
1820             extract_add_complete_field(p, zm);
1821         else
1822             extract_add_incomplete_field(p, zm);
1823     }
1824 }
1825
1826 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1827                                       void *buf, size_t sz)
1828 {
1829     ZebraHandle zh = (ZebraHandle) p->handle;
1830
1831     xfree(zh->store_data_buf);
1832     zh->store_data_buf = 0;
1833     zh->store_data_size = 0;
1834     if (buf && sz)
1835     {
1836         zh->store_data_buf = xmalloc(sz);
1837         zh->store_data_size = sz;
1838         memcpy(zh->store_data_buf, buf, sz);
1839     }
1840 }
1841
1842 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1843 {
1844     ZebraHandle zh = (ZebraHandle) p->handle;
1845     xfree(zh->store_data_buf);
1846     zh->store_data_buf = 0;
1847     zh->store_data_size = 0;
1848     p->setStoreData = extract_set_store_data_cb;
1849 }
1850
1851 static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid)
1852 {
1853     ZebraHandle zh = (ZebraHandle) p->handle;
1854     zebraExplain_addSchema(zh->reg->zei, oid);
1855 }
1856
1857 void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
1858                              int cmd, zebra_rec_keys_t reckeys)
1859 {
1860 #if 0
1861     yaz_log(YLOG_LOG, "extract_flush_sort_keys cmd=%d sysno=" ZINT_FORMAT,
1862             cmd, sysno);
1863     extract_rec_keys_log(zh, cmd, reckeys, YLOG_LOG);
1864 #endif
1865
1866     if (zebra_rec_keys_rewind(reckeys))
1867     {
1868         zebra_sort_index_t si = zh->reg->sort_index;
1869         size_t slen;
1870         const char *str;
1871         struct it_key key_in;
1872
1873         zebra_sort_sysno(si, sysno);
1874
1875         while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1876         {
1877             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1878
1879             zebra_sort_type(si, ord);
1880             if (cmd == 1)
1881                 zebra_sort_add(si, str, slen);
1882             else
1883                 zebra_sort_delete(si);
1884         }
1885     }
1886 }
1887
1888 /*
1889  * Local variables:
1890  * c-basic-offset: 4
1891  * indent-tabs-mode: nil
1892  * End:
1893  * vim: shiftwidth=4 tabstop=8 expandtab
1894  */
1895